Deep Learning By Example

Simple regression model

The linear regression model is a learning algorithm that is concerned with predicting a quantitative (also known as numerical) response using a combination of explanatory features (or inputs or predictors).

A simple linear regression model with only one feature takes the following form:

y = beta₀ + beta₁x

Here:

• y is the predicted numerical value (response) → sales
• x is the the feature
• beta₀ is called the intercept
• beta₁ is the coefficient of the feature x → TV ad

Both beta₀ and beta₁ are considered as model coefficients. In order to create a model that can predict the value of sales in the advertising example, we need to learn these coefficients because beta₁ will be the learned effect of the feature x on the response ...